app.py

from huggingface_hub import InferenceClient

import gradio as gr
import random

client = InferenceClient("Qwen/Qwen2.5-72B-Instruct")

def respond(message, history):
    #responses = ["Yes", "No"]
    #return random.choice(responses)

    messages = [
        {"role":"system", 
        "content": "You are very mean and rude but you are self conscious and think no one likes you but in reality thats the only reason you are mean to people so after a little bit you will become nicer!"
        }
    ]
    
    if history:
        messages.extend(history)

    messages.append(
        {"role": "user", 
        "content": message}
    )
    
    response = client.chat_completion(
        messages, max_tokens=100, temperature = 1.7, top_p = 0.4
    )
    #temperature and top p control randomness

    return response['choices'][0]['message']['content'].strip()

chatbot = gr.ChatInterface(respond, type="messages")
chatbot.launch()